WASHINGTON. In the midst of images depicting the destruction in Gaza, certain ones have sparked genuine horror: images of bloodied, abandoned infants. These visuals, circulating widely online since the beginning of the war, are, in fact, deepfakes crafted using artificial intelligence. Closer inspection reveals subtle clues, such as oddly curled fingers or eyes with an unnatural light, exposing the signs of digital manipulation.
Despite the digital deception, the outrage these images aim to incite is undeniably authentic. The Israel-Hamas conflict has vividly demonstrated the potential of AI as a propaganda tool, generating lifelike images of devastation. Throughout the war, digitally altered images disseminated on social media have been employed to make false claims about casualty responsibility or to deceive the public about non-existent atrocities.
While most false claims related to the war didn’t necessitate AI and originated from conventional sources, technological advancements, with little oversight, are becoming more frequent. This highlights the stark reality that AI has the potential to evolve into another form of weaponry, offering a glimpse of what may occur in future conflicts, elections, and significant events.
Jean-Claude Goldenstein, CEO of CREOpoint, a tech company utilizing AI to assess online claims, warns, “It’s going to get worse—a lot worse—before it gets better.” The company has compiled a database of the most viral deepfakes emerging from Gaza. Goldenstein emphasizes that with generative AI, involving pictures, videos, and audio, escalation beyond current levels is anticipated.
Some instances involve repurposing photos from other conflicts or disasters, presenting them as recent. In other cases, generative AI programs have generated images from scratch, like a viral image of a crying baby amidst bombing wreckage during the conflict’s early days.
AI-generated content also includes videos depicting purported Israeli missile strikes, tanks moving through ruined neighborhoods, and families searching through rubble for survivors.
Many of these fakes aim to evoke a strong emotional reaction by featuring the bodies of babies, children, or families. Propagandists are adept at exploiting people’s deepest impulses and anxieties, emphasizing the emotional impact to further disseminate disinformation.
Imran Ahmed, CEO of the Center for Countering Digital Hate, notes that whether it’s a deepfake or an actual image, the emotional impact on the viewer is the same. The more abhorrent the image, the more likely it is to be remembered and shared, inadvertently spreading disinformation.
Similar deceptive AI-generated content emerged after Russia’s invasion of Ukraine in 2022, indicating the persistent nature of easily debunked misinformation.
With major elections approaching in various countries, the risk of AI and social media spreading lies has raised concerns. Lawmakers in Washington, including U.S. Rep. Gerry Connolly, emphasize the need for investment in AI tools to counteract other AI.
Several startup tech firms worldwide are developing programs to detect deepfakes, affix watermarks to images for origin verification, or scan text to identify specious claims introduced by AI.
As technology advances, experts emphasize the necessity for improved regulations, voluntary industry standards, and substantial investments in digital literacy programs to address the challenges posed by AI disinformation effectively. The goal is not just detection but a comprehensive solution to tackle the broader political and social implications of AI manipulation.
AP contributed to this story.
